Neuropathological Staging and AD
The neuropathological staging proposed by Braak (Braak, H et al. (1991), Acta. Neuropathol. 82, 239-259) provides the best available definition of progression of relatively pure neurofibrillary degeneration of the Alzheimer-type which is diagnostic of AD (Wischik et al. (2000), “Neurobiology of Alzheimer's Disease”, Eds. Dawbarn et al., The Molecular and Cellular Neurobiology Series, Bios Scientific Publishers, Oxford). This staging is shown schematically in terms of brain region in FIG. 2B, and is based on a regular regional hierarchy of neurofibrillary tangle (NFT) distribution. Regions of the brain which appear earlier in the hierarchy have both more tangles and are affected in less severe cases than those later in the list.
Relationship Between AD, Clinical Dementia and Neuropathological Staging
The provision of an effective pre-mortem assessment of Braak Stage would be of use in the assessment and treatment of AD, for which the differential includes Lewy Body dementia, Parkinson's disease, various forms of fronto-temporal and cortico-basal degeneration, progressive supranuclear palsy and a range of rare neurological syndromes.
The original model proposed by Braak was essentially qualitative in nature, and was not linked to any implications about the threshold for development of clinical dementia and symptoms.
In terms of the appearance of clinical dementia by DSM-IV criteria, this corresponds statistically to the transition between Braak stages 3 and 4 (FIG. 2c). The DSM-IV criteria (Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, American Psychiatric Association, American Psychiatric Press, Washington D.C. (1994)) for the definition of dementia equate to an MMSE (Mini-Mental State Examination) cut-off point of about 18, and corresponds to a dementia prevalence of about 5% of the population over-65 years old (over-65's represent about 17% of the total population).
Gertz et al. ((1996) Eur. Arch. Psychiatry Clin. Neurosci. 246, 132-6)) studied cases followed from general practice to post-mortem, which were rigorously characterised at the clinical level using CAMDEX (Roth et al, 1988, “The Cambridge Examination for Mental Disorders of the Elderly (CAMDEX)” Cambridge University Press). These were staged post-mortem by the criteria of Braak and, after excluding all cases with any degree of vascular pathology found post-mortem, there remained uncertainty in about one third of cases at the point of transition. That is, about one third of cases with a clinical diagnosis of AD are actually at early Braak stages (stages 1-3), have vascular pathology, or have concomitant Lewy body pathology. Thus there exists a high degree of uncertainty, even in the best practice research setting. The predominant neuropathological substrate that is actually present when a routine clinical diagnosis of AD is made is even more uncertain.
It has recently been reported that a molecule (PDDNP, 2-(1-{6-[(2-[18F]fluoroethyl)(methyl)amino]-2-naphthyl}ethylidene)malononitrile) demonstrates increased relative retention time (RRT) in medial temporal lobe brain regions (hippocampus, entorhinal cortex and amygdala) following injection and PET imaging in cases with a clinical and neuroradiological diagnosis of Alzheimer's disease (Shoghi-Jadid et al., Am. J. Geriatr. Psychiatr. 2002, 10:24-35).
Although binding to NFTs and amyloid plaques is discussed, no binding to NFTs is shown, although the compound does bind with high affinity to synthetic beta-amyloid fibrils in vitro.
When cases were matched for corresponding disease severity by MMSE score with a neuropathological case series in which vascular pathology was excluded, the RRT values reported by Shoghi-Jadid et al. were found to correlate with beta-amyloid plaque counts but not with measures of neurofibrillary tangle pathology, as shown below.
Spearman Rank Correlation Coefficients:
MTL APGlobal APMTL NFTGlobal NFTRRT0.665**0.654**0.2440.189p<0.01<0.01>0.1>0.1
Pearson Correlation Coefficients:
MTL APGlobal APMTL NFTGlobal NFTRRT0.602*0.596*0.2660.275p<0.05<0.05>0.3>0.3
Where the parameters are defined as follows:
MTL APmedial temporal lobe amyloid plaquesGlobal APaverage amyloid plaque load in 12 brain regionsMTL NFTmedial temporal lobe neurofibrillary tanglesGlobal NFTaverage NFT load in 12 brain regions
However, beta-amyloid deposition is known to discriminate poorly between normal aging and Alzheimer's disease (see FIG. 2d herein), and beta-amyloid pathology does not provide a sound basis for neuropathological staging (Braak and Braak, 1991). Therefore, FDDNP-RRT does not provide a method for in vivo neuropathological staging of Alzheimer's disease.
Although it is possible that further refinement in clinical methods with particular reference to more specific neuropsychological indicators (e.g. split attention tasks, delayed matching to sample, etc.) may improve the accuracy of clinical diagnosis, an essential problem is to develop methods for the direct measurement of underlying neuropathology during life, in particular the extent of neurofibrillary degeneration of the Alzheimer-type.
Progression of Neurofibrillary Degeneration and Tau
As described above, the tau-based pathology of AD is a major feature of the phenotype. It is highly correlated with extent of neuronal destruction (reviewed in Wischik et al. (2000) loc cit).
On a cellular basis, the formation of NFTs from Tau is believed to proceed as follows. In the course of their formation and accumulation, paired helical filaments (PHFs) first assemble as filaments within the cytoplasm, probably from early tau oligomers which become truncated prior to, and in the course of, PHF assembly (Refs 26, 27). They then go on to form classical intracellular NFTs. In this state, PHFs consist of a core of truncated tau and a fuzzy outer coat containing full-length tau (Wischik et al. (2000) loc. cit.). The assembly process is exponential, consuming the cellular pool of normal functional tau and inducing new tau synthesis to make up the deficit (Ref 29). Eventually functional impairment of the neurone progresses to the point of cell death, leaving behind an extracellular NFT. Cell death is highly correlated with the number of extracellular NFTs (Bondareff, W. et al. (1993) Arc. Gen. Psychiatry 50: 350-6). As the outer neuronal membrane is damaged and NFTs are extruded into the extracellular space, there is progressive loss of the fuzzy outer coat of the neurone with corresponding loss of N-terminal tau immunoreactivity, but preservation of tau immunoreactivity associated with the PHF core (FIG. 3; Ref 30).
In the process of aggregation, tau protein undergoes a conformational change in the repeat domain associated with a half-repeat phase-shift (Refs 32, 33). This creates a proteolytically-stable fragment which is identical to that found in the core of the paired helical filaments (PHFs) which constitute the neurofibrillary tangles characteristic of AD. By analogy with other protein aggregation systems, the process most likely involves an alpha-helix to beta-strand change in conformation (reviewed in Wischik et al. (2000) loc. cit.).
Generally speaking therefore, the aggregation of tau can be considered in 3 stages: intracellular oligomers; intracellular filaments (stage 1 of FIG. 3); extracellular filaments (stages 2 & 3 of FIG. 3).
However, to date, no definitive correlation has been established between these stages, which occur at the cellular level, and possibly at different rates and probabilities in different regions in the brain, and the progression of pathology according to the defined hierarchical system of Braak and Braak which, as discussed above, is the best available definition of progression of relatively pure neurofibrillary degeneration.
Invasive Methods for Assessing AD
Lumbar-puncture CSF measurements enable discrimination between AD and controls, and between AD and other neurological disorders, but lumbar-puncture is more invasive then nuclear medicine-based approaches, and carries a higher risk (Refs. 17 to 21). EEG-neurological diagnosis has also been developed (Refs 22-25), but in this regard there remains a need for cheap instrumentation which can be used at the point of clinician contact.
Neurofibrillary Degeneration Via Brain Atrophy —SPECT and PET
Numerous studies have shown that global brain atrophy and specific medial temporal lobe atrophy, particularly of the hippocampus, are closely linked to underlying neurofibrillary degeneration of the Alzheimer-type, and are valuable in the early diagnosis of AD (Refs 1-8).
However, although the diagnosis of AD by monitoring global brain atrophy represents a methodology which can be made to work in a research setting, there are difficulties in defining and measuring atrophy in specific brain regions, and likewise in the measurement of global neocortical atrophy. In any case, a diagnosis based on detectable atrophy may come too late for effective treatment.
There have been advances in diagnostic methodology in recent years, following the identification of diagnostic features in SPECT scans (Refs 9-12; characteristic patterns of perfusion defect detected by HMPAO SPECT), PET scans (Refs 13-15; metabolic defect detected by glucose metabolism profile) and MRI scans (Ref 16; global brain atrophy, specific patterns of lobar atrophy). Of these, the most generally accessible are MRI and SPECT, since PET is for the present time limited to centres which have local specialised cyclotron and radiochemistry capability for the preparation of short half-life injectable radioligands (Aberdeen, London, Cambridge in UK). Notably, the characteristic early stage temporo-parietal perfusion defect detected by HMPAO SPECT in patients with AD corresponds very closely to the pattern of tau pathology which can be detected biochemically (FIG. 1). The biochemical changes precede overt neurofibrillary degeneration as seen by the appearance of NFTs (FIG. 2; Mukaetova-Ladinska et al., 2000 Am. J. Pathol. Vol. 157, No. 2, 623-636).
However, although MRI and SPECT scans are useful for detecting specific patterns of perfusion defects characteristic of AD, discrimination between different neuropathological stages, or between AD and other types of dementia, is difficult.
For instance, SPECT is useful for detection of a specific pattern of bilateral temporo-parietal perfusion defect that is characteristic of AD (Refs 9-11), which can be useful even at very early stage disease. However, SPECT changes discriminate neuropathological stages poorly (Ref 12). Furthermore, discrimination between AD and Lewy Body dementia is difficult. Both have a bilateral temporo-parietal perfusion defect, but only in the latter does an occipital perfusion defect tend to be present. The same patterns of defect can be demonstrated using PET measurement of glucose metabolism (Refs 1315), but the problem of distinguishing Lewy Body dementia persists in this modality.
Thus, as can be inferred from data in Ref 12, the probability of successful SPECT detection of cases at Braak stages 1&2 is 50%, and at stages 3&4 is 60%. It is only at stages 5&6 that 95% of cases become SPECT-positive. Conversely, cases detected as SPECT-positive could be at stages 1&2 (20%), 3&4 (20%), or 5&6 (60%). SPECT will therefore fail to detect 40-50% of the pre-stage 4 target population for early diagnosis and therapeutic intervention. In a further study (data not shown) it was shown that overall agreement between SPECT diagnosis and clinical diagnosis was of the order of 50%.
Thus, in developing a treatment aimed specifically at preventing neurofibrillary degeneration of the Alzheimer-type, there is a critical need to develop, in parallel, non-invasive means of selecting patients for treatment, and monitoring their response to the treatment, according to a defined and reproducible definition of disease progression.